Hi Yanling,

To get maximum throughput you need to use VBO's and break your
geometry up into chunks of say around 1-10,000 vertices each.

Robert.

On 7/29/06, Yanling Liu <[EMAIL PROTECTED]> wrote:
Hi, does any of you guys achieve the maximum throughput about triangle
rendering of your video card? I am asking this is because my program running
on a NVIDIA Quadro FX 3400 + Xeon 3.6G single CPU could render a human brain
model having 3 million triangles at 20 fps. According to NVIDIA's website, a
Quadro FX 3450 has theory maximum throughput 150 million triangles per
second. (I guess 3400 has no big performance difference with 3450). So the
card could process about 7 million triangles per frame at 20 fps rendering
speed. Apparently my program is far away from my video card's maximum
throughput. I already used space partition to speedup and I have no idea
where is the bottle neck. Could anyone help me to know the difference
between the theroy maximum rendering throughput and practical throughput
when we rendering something? Say if the card could process 150 million
triangles, what is the typical practice performance? 50 million triangles?
100 million triangles? could we achieve the maximum throughtput by some
special optimization?

Two more things need to be mentioned: 1. I did not use triangle strip.
Instead, I have used DrawArrays using PrimitiveSet::TRIANGLES. Also there is
not much difference if I switch to DrawElementsUInt. 2. displaylist has been
used in my program.

Thanks ahead!
Yanling

_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/


_______________________________________________
osg-users mailing list
[email protected]
http://openscenegraph.net/mailman/listinfo/osg-users
http://www.openscenegraph.org/

Reply via email to